GitLab

If run-test leaves cruft in $ANDROID_PRODUCT_OUT, flashall -w will
make the files on the device owned by the system user. This will cause
run-test 071-dexfile to fail when run as root.

Change-Id: I071a1d52cbbefd0e75556706b473260cb6832fd4

Fixes failure in libcore.java.io.OldAndroidSerializationTest#testSerialization

Change-Id: I13187f081ea87402be88d59a48ec3890505342ae

Also make .art the primary build rule, not the .oat. The
build/core/dex_preopt.mk rules depend on the .art files, and without
this, unnecessary rebuilds can happen when the .art file is
regenerated.

Change-Id: I0b116d6975916b9140e7ef095f8b0ec994c8e50f

If-conversion of the arm monitor enter/exit logic to avoid
mispredicted forward branch.  Small, but measureable improvement
in synchronization benchmark.

Change-Id: I88154a9690258099ec1d6de6297e9a7bade97d88

Change-Id: I160bc78ef9000364f657a9972dd8c648ca2f53aa

Seeing new instances of this C-ism go in makes me a sad panda.

Change-Id: Ie3dd414b8b5e57a4164e88eb2d8559545569628d

Change-Id: I288337af4c70716709217ff2d21050ba5f858807

commit a433b2e8 just removed this code.

Change-Id: I9561b40c659ca62527e4dbfce33249ed34731b6a

Change-Id: I09c76b8fede9e926519b206010ba6c7ab44c3216

The compilers had drifted a bit while I was away.

Change-Id: I938a5992e8e100f512db4c7934ac4b3ec80e8138

This builds three separate compilers and dynamically links with the right one
at runtime.

Change-Id: I59d22b9884f41de733c09f97e29ee290236d5f4b

Helpers or utilities such as FindFieldFast are still in
runtime_support_common. But LLVM has different stacks, no callee-saves
and different exception mechanisms. Those won't belong to
runtime_support_common.

Change-Id: Ibc43d7d53316941481a6a21815ddefdc1e6bdde6

The bug was due to a refactoring error last night. We will also take out
_artInvokeCommon and other confusing de-dup in the next CL. But I want
to fix the crash right away.

Change-Id: Ic3fcb0bbaaca44cf368f79994a0cdcfac5fdfc8b

Fix initialization bug in thread.cc.

Change-Id: Ibdc81d6f5c8b1a245044bf901003179167a5d4f1

Change-Id: I59df076e148b571d253df2aa3bb852fc032c9714

Change-Id: Ieac2be41efcb57923de6f1dedad94fab0803812d

Change-Id: Id543e71c01b65b40abdfd658f6969b3d62c6e04e

Change-Id: I3dd20bf3e9da85a99abdee9994d1d02cbbda86cf

This version tried to use std::mismatch instead of a loop.

Change-Id: I463bef4a36dd91573369754db2f32f3d45fc53dc

A regression in voice search was seen. Disable for now. Add a few
assertions to aid future diagnostics.

Change-Id: I635b1126b3e45293489b6b7ad54d838773ae7030

Change-Id: Ib05030d29a2d9c1ee08413f24ca9c7c2082bf72f

Addressed the issue that we shouldn't maintain 2 copies of almost the
same code.

Change-Id: I212656f65e6c47e1167a193c1024100ae97f0028

LIR operand 0 is always an offset for a branch. This is clear in
conditional branches that are binary and have the 2nd operand as the
condition codes of the branch. This changes unconditional branches to be
unary and therefore more intention revealing that the 1st operand will
be used by the assembler to hold an offset.

A << operator for OpKind allows easy pretty printing.

Change-Id: I933b8e0bf43f5be3eff13f93c3fc1539ae526840

First stab at translating the Arm stubs to MIPS code.

Change-Id: I0cff280c760b6355e2ede85434752efe1e425c58

The decoder was not populating register vA and vB.

Change-Id: I5a162f308ef919f4c98efe9bbdf8eb263906d01a

There are a few "safe" optimizations in the compiler - removing
register copies where source and target are the same, deleting
branches to the next instruction, etc.  One of the redundant
branch optimizations, however, was incorrect and resulted in
a good branch being deleted.  This one showed up in the debug
build, and resulted in a failure to do a suspend check (because
the branch to the suspend check was deleted).

I had hoped that this but might also be the case of some
other unexpected failures, but unfortunately I was only able
to trigger it when doing a "codegen for debug" build.

The source of the bug was a confusion around 16 v/ 32-bit
unconditional branch encodings.  For a 32-bit unconditional
branch, going to the next instruction means an displacement
of zero.  However, for 16-bit branches, the next instruction
is represented by a displacement of -1.

To help track down this sort of thing in the future, this CL
also adds a new optimization disable flag: kSafeOptimizations.
This will allow us to really turn off all optimizations for A/B
testing.

Also in this CL we are re-enabling the ability to promote argument
registers and improving somewhat the code sequence for suspend
check when debug is enabled.

Change-Id: Ib6b202746eac751cab3b4609805a389c18cb67b2

Yesterday we added a static guard, which is removed from this commit.
Using llvm_shutdown_obj will avoid the ordering problem.

Change-Id: Iaa2e4293373c849f6b07082fb70b49ff0e805033

Change-Id: Idb6165ba4525293b5ac8d65a5f37240ef6bc3483